Journal of the American Mosquito Control Association, 13(1): 13-17, 1997
METHODS OF TESTING AND ANALYZING EXCITO-REPELLENCY
RESPONSES OF MALARIA VECTORS TO INSECTICIDES 1
DONALD R. ROBERTS, 2 THEERAPHAP CHAREONVIRIYAPHAR 3
HAROLD H. HARLAN 4 and PAUL HSHIEH 2
ABSTRACT. A new test system that includes an excito-repellency test box, test procedures, and statistical
treatment of data is described. The method consists of enclosing 25 mosquitoes in an exposure chamber lined
with insecticide-treated or untreated (control) test papers. Each chamber has a single portal for mosquitoes to
escape to a receiving cage, and numbers escaping are manually recorded at 1-min intervals. The exposure
chamber accommodates a screened, 2nd chamber that, when placed in the exposure chamber, prevents the
mosquitoes from making physical contact with test papers. A full assay utilized one exposure chamber that
permits physical contact with insecticide-treated papers, one chamber that permits physical contact with control
papers, one chamber that prevents physical contact with insecticide-treated papers, and a 4th chamber that
prevents contact with control papers. After insecticide exposure, test populations are held for observations on
24-h mortalities. A survival analysis approach is described for estimating mosquito escape rates and for com-
paring differences in mosquito escape rates, with or without physical contact with insecticide, among populations,
insecticides, and doses of insecticide.
Assays for evaluating behavioral responses of
malaria vectors to insecticide residues have been
reviewed by Muirhead-Thomson (1960), Coluzzi
(1963), Busvine (1964), and Elliott (1972). The test
of greatest value for studies of insecticide avoid-
ance was described by Coluzzi (1963) as a box with
slits for escaping. Such a box was described by
Rachou et al. (1973) and is referred to as the excito-
repellency test box. Similar excito-repellency test
boxes are described by Rachou et al. (1973), Charl-
wood and Paraluppi (1978), Roberts et al. (1984),
Rozendaal et al. (1989), and Evans (1993). In ex-
cito-repellency tests, mosquitoes are released inside
a box lined with sprayed paper. Outlets in the form
of out-projecting baffles permit the mosquitoes to
escape into 2 separate cages. The baffles prevent
the mosquitoes from reentering the box and the
numbers escaping are counted by time postrelease.
The difficulties of working with test boxes were
described by Roberts et al. (1984). Major problems
relate to the difficulties in introducing specimens
into the boxes, removing live specimens at the end
of test periods, and providing a standardized insec-
ticide dose. The lack of an appropriate method of
data analysis has been another shortcoming of the
test method. Earlier methods did not test for behav-
1 This research was supported by grant R087EK from
the Uniformed Services University of the Health Sciences,
Bethesda, MD. The views of the authors do not purport
to reflect the positions of the U.S. Department of Defense
or the Uniformed Services University of the Health Sci-
2 Department of Preventive Medicine and Biometrics,
Uniformed Services University of the Health Sciences,
4301 Jones Bridge Road, Bethesda, MD 20814-4799.
3 Kampheangsaen Campus, Kasetsart University, Fac-
ulty of Liberal Arts & Sciences, Nakhon Prathom 73140,
4 Maple Hill Lane, Crownsville, MD 21032-1062.
ioral responses without physical contact with insec-
Described herein are improved boxes for testing
behavioral responses of adult Anopheles mosqui-
toes with or without physical contact with insecti-
cide residues. Survival analysis methods are de-
scribed for the statistical treatment of test data.
MATERIALS AND METHODS
The test method consists of enclosing 25 mos-
quitoes in an exposure chamber lined with insecti-
cide-treated or untreated (control) papers. Each ex-
posure chamber has a single portal for mosquitoes
to escape to a receiving cage. The exposure cham-
ber accommodates a screened, 2nd chamber (inner
chamber) that, when placed in the first chamber,
prevents the mosquitoes from making physical con-
tact with test papers. Under test conditions, mos-
quitoes are enclosed within the exposure chamber
and the only source of light comes from the exit
portal. A full assay consists of 4 exposure chambers
of 2 treatment chambers and 2 control chambers,
as shown in Table 1 . Treatment chambers are lined
with test papers impregnated with insecticide and
an oil-based carrier. Control chambers are lined
with papers impregnated with carrier alone. One
treatment chamber permits tarsal contact with in-
secticide. The second treatment chamber includes
the inner chamber, so mosquitoes cannot make tar-
sal contact with insecticide. For brevity, tests with
or without the inner chambers, for either treatment
or control papers, are referred to as contact trials
(no inner chamber) or noncontact trials (with an
Components of the excito-repellency chambers
are illustrated and numbered in Fig. 1. Except for
a inner panel (No. 1), the exposure chamber is con-
structed of metal and can be chemically cleaned.
The exposure chamber (No. 4) is constructed of
stainless steel and each chamber is 34 X 32 X 32
Journal of the American Mosquito Control Association
Vol. 13, No. 1
Table 1 . Test conditions for evaluating behavioral
responses of malaria vectors to insecticide residues.
Papers lining the
With or without
with test papers
cm. The front panel is 32 X 32 cm and is equipped
with an escape portal (No. 6). The escape portal is
an outward projecting funnel (exit funnel), 14.75
cm at its base. The top and bottom of the exit fun-
nel are 14 cm long and converge, leaving a 1.50-
cm-wide opening (a horizontal slit) through which
the mosquitoes can escape from the exposure
chamber. The back of the exposure chamber is a
hinged metal door (No. 5) that closes tightly. The
exposure chamber is also equipped with an inner
removable rear panel (No. 1). This panel fits inside
the back of the exposure chamber, abuts 4 small
flanges inside the chamber, and serves to imprison
31 X 30.5 cm.
2: Dental dam, sealed
port (15.5 cm in dia.) for putting
specimens inside the chamber.
3: Screened inner chamber,
28.5 X 28.5 X 29 cm.
4: Stainless steel exposure
(outer) chamber, 34 X 32 X 32 cm.
5: Hinged, stainless steel rear door.
6: Stainless steel escape louvre,
slit was 1 .5 cm wide.
Fig. 1. An excito-repellency test box for the study of behavioral responses of mosquitoes to insecticides.
the test population inside the exposure chamber.
Plexiglas® is used for the inner rear panel so mos-
quitoes can be observed inside the chamber. The
Plexiglas panel is equipped with a large round hole
(15.5 cm in diameter) that is sealed with a split
piece of dental dam (No. 2). This sealed opening is
used for placing mosquitoes inside and for remov-
ing mosquitoes from the chamber. The 2 rear panels
fulfill several requirements. First, the test popula-
tion must be in darkness so imprisoned specimens
can orient on light filtering through the escape fun-
nel; thus the chamber needs to be solid and non-
transparent. This requirement is fulfilled by closing
the rear metal door at the start of each test. Second,
the investigator needs to see inside the chamber to
check for dead versus live specimens, both pre- and
posttesting, and to remove live specimens at the
end of the test. The ability to see inside the chamber
is fulfilled by using transparent Plexiglas for the
inner rear panel. Third, a self-sealing portal is need-
ed for placing a test population inside the chamber
and for removing specimens from the chamber at
the end of each test. This requirement is fulfilled
by using a split dental dam seal on the 15.5-cm-
diameter opening in the Plexiglas rear panel.
The frame of the inner chamber (No. 3) is con-
structed of 0.62 X 0.62-cm aluminum beams. The
structure of each chamber is 28.5 X 28.5 X 29 cm
and the inner surface of each is covered with metal
screening. A fine mesh metal screen, 52 cells per
inch, covers the top, bottom, and 2 side walls of
the inner chamber. The inner chamber is open end-
ed, with 0.62-cm rubber gaskets on the front and
back beams. When placed in the outer chamber, the
front gasket seals small gaps between the front
stainless steel panel and the inner chamber. Like-
wise, the rear gasket seals gaps between the Plexi-
glas panel and the inner chamber. The inner screen
surface is no closer than 0.62 cm from the surface
of test papers, and it prevents mosquitoes from
making tarsal contact with the surface of test pa-
The receiving cage is a one-gal (1.6-liter) ice
cream carton with a screened top. The cage fits over
the outward projecting exit funnel. A section of or-
thopedic stocking is attached to an opening in the
side of the carton. The funnel of the exposure
chamber is inserted through the stocking and
through the opening of the receiving cage.
Test papers lining the exposure chambers are first
clipped, with metal paper clips, to large sheets of
clean white typing paper. The large papers are taped
together in a ribbon effect. Then, with test papers
attached, the ribbon of paper is placed against the
sides, top and bottom of the exposure chamber. Pa-
pers are secured to the walls by pairing a small
magnet on outside of the wall with a paper clip on
the inside wall, the paper is secured by attraction
between the magnets and paper clips. Test papers
are not positioned on the front or back of the ex-
A full test requires 4 groups of 25 mosquitoes
(test population) each. With 2 investigators, test
populations can be introduced into each of the 4
exposure chambers in approximately 1 min. Before
mosquitoes are introduced into the exposure cham-
bers, exit funnels are sealed with Styrofoam® in-
serts. A 3 -min rest period has been used to permit
mosquitoes to adjust to test chamber conditions in
other test procedures (Busvine 1964); therefore, a
3-min interval is used in the present procedure. Af-
ter 3 min, the Styrofoam insert is removed from
each of the escape funnels to initiate the observa-
tion period. Numbers escaping from exposure
chambers to receiving cages are recorded manually
at 1-min intervals; after 5 min of observation, re-
ceiving cages are replaced with clean cages. The
exchange of receiving cages facilitates the accurate
counting of numbers escaping for each time inter-
A survival analysis approach is used to estimate
the rates of mosquitoes escaping from chambers. In
the excito-repellency test, there are only 2 possible
outcomes for a specimen: it will either escape or
not escape from the exposure chamber. Binary test
data are optimized for survival analysis techniques
by only working with counts of specimens that do
not escape. However, an estimate of escape rate or
probability of escape is obtained by subtracting
from one the estimated rate or estimated probability
of remaining in the exposure chamber. These sta-
tistically defined estimates for 1-min observation
periods can be used to compare differences in mos-
quito escape rates among populations, insecticides,
and concentrations (doses) of insecticides, in either
contact or noncontact trials. The analytical results
can be presented either as proportions escaping or
proportions remaining in exposure chambers.
In this analysis, mosquitoes that escape are treat-
ed as "deaths" and those remaining in the exposure
chamber from one minute to the next as "surviv-
als." Specimens in the exposure chamber at the end
of the test are treated as "censored." In survival
analysis terminology, the survival time of a speci-
men is thought to be censored when the end point
of interest (in our case, escape from the exposure
chamber) has not been observed for that specimen
(Lee 1992, Collett 1994). Time (min) for 50% and
90% of the test population to escape is estimated
with the life table method, and these estimates are
used as "escape time" summary statistics (ET^ and
The log-rank method is used to compare patterns
of escape behavior (analogous to survival curves).
This test is designed to detect differences between
survival curves that result when the death (or es-
cape) rate in one group is consistently higher than
the corresponding rate in a 2nd group and the ratio
of these 2 rates is consistent over time (in survival
analysis, this is also called the proportional hazard
rate). With excito-repellency data, the basic idea
underlying the log-rank test involves examining es-
Journal of the American Mosquito Control Association
Vol. 13, No. 1
cape observations by 1-min intervals. To test the
null hypothesis, we calculate the observed escape
and expected escape in each 1-min interval. The
data are analyzed by use of tabular data presenting
columns for time, number observed to escape,
number expected to escape, and difference between
observed and expected. We then combine the tab-
ular data for each test to give an overall measure
of the deviation of the observed escape values from
their expected values by each 1-min test interval.
The log-rank method was proposed by Mantel and
Haenzel (1959); it is also called the Mantel-Cox
and Peto-Haenzel methods (Mantel and Haenzel
The log-rank test has a chi-square distribution
with k degrees of freedom, where k is the number
of groups- 1. A statistical software package, STA-
TA® 5 , can be used for this analysis to test for dif-
ferences among or between populations, dose lev-
els, and insecticides.
RESULTS AND DISCUSSION
The World Health Organization's (WHO) rec-
ommended tests of malaria vectors for behavioral
responses to insecticides do not discriminate be-
tween contact versus noncontact stimulation (WHO
1975). The tests are based on the concept that ma-
laria vectors respond to insecticides only after
physical contact with the chemical and this concept
As measured in the excito-repellency test (Char-
eon viriyaphap et al. 1997), noncontact repellency is
not as quick or pronounced a behavioral response
as contact irritancy. In the field, most mosquitoes
stimulated to prematurely exit houses will probably
do so only after physical contact with insecticide
residues. However, the mosquito must first enter the
house before it can make contact with insecticide.
Except for a specimen that enters and then exits the
house in pursuit of a host, the stimuli for the 3
behavioral acts of entering, resting indoors, and
eventually exiting the house are different.
Noncontact repellency probably exerts its most
powerful influence by preventing mosquitoes from
entering houses. This repellency action has been
documented in several field studies against several
vector species (Roberts and Andre 1994). As ex-
amples, Roberts and Alecrim (1991) showed with
experimental houses that Anopheles darlingi Root
females practically stopped entering a house after
it was sprayed with DDT, approaching a 100% re-
duction in indoor biting. Smith and Webley (1968)
showed a 60-70% reduction in house entering by
Anopheles gambiae Giles females after an experi-
mental house was sprayed with DDT. Shalaby
(1966) showed that, in comparison with a control
5 STATA* statistical software was provided by Stata
Corporation, 702 University Drive East, College Station,
house, 75% fewer Anopheles culicifacies Giles
specimens were collected inside a DDT-sprayed
house, even with all surfaces screened to preclude
physical contact with DDT. With this background,
it is clear that the excito-repellency test may pro-
vide a direct measure of vector responsiveness to
the irritant affects of insecticides, and the results
might be predictive of actions inside of houses un-
der field conditions. However, the excito-repellency
test is perhaps less informative of insecticidal im-
pact on the act of entering a sprayed house.
Behavioral responses of vectors to insecticides
are important, but generally neglected areas of
study. Progress in understanding the importance of
insecticide avoidance behaviors has been impeded
by the lack of acceptable test systems. To date, no
WHO-recommended test methods discriminate be-
tween contact versus noncontact insecticide-in-
duced behaviors. No tests are easily conducted or,
excluding the excito-repellency test box, provide a
powerful and reproducible result. Additionally, no
test data are amenable to sophisticated statistical
analyses. The excito-repellency test box, the test,
and data analysis methods (test system) described
in this report were designed to resolve some of the
problems identified by Roberts et al. (1984). The
test system has already been used in an extensive
study of behavioral responses of different Anoph-
eles albimanus Wiedemann populations to DDT,
permethrin, and deltamethrin (Chareonviriyaphap
et al. 1997). Using this test system in combination
with susceptibility, isozyme, and esterase tests,
Chareonviriyaphap et al. (1997) obtained no evi-
dence of relationships between physiological resis-
tance and behavioral responses of An. albimanus
females to insecticides.
The excito-repellency test system described in
this report offers the following desirable attributes:
1 . Exposure chambers are constructed of metal and
can be chemically cleaned for use with different
doses and types of insecticides.
2. Screened inserts provide a capability to test be-
havioral responses without physical contact with
3. Mosquitoes are easily transferred to the expo-
sure chambers and are easily removed from the
chambers after the test is complete.
4. Counts by 1-min intervals are sensitive to rapid
behavioral responses to insecticides.
5. Highly reproducible test results are obtained
(Chareonviriyaphap et al. 1997).
6. The survival analysis method is a robust treat-
ment of data that minimizes the loss of infor-
7. Comparative summary statistics in the form of
escape times for 50 and 90% (ET 50 and EX*,) of
test specimens to escape exposure chambers can
8. Specimens that escape at different time intervals
and specimens that remain in the exposure
chamber throughout the test can be held and
scored for 24-h mortalities.
In our studies (Chareonviriyaphap et al. 1997),
we employed test papers that were prepared ac-
cording to WHO specifications by the United States
Army Center for Health Promotion and Preventive
Medicine, Aberdeen Proving Ground, MD. Test pa-
pers are also available from the WHO.
This research was supported by an intramural
grant (R087EK) from the Uniformed Services Uni-
versity of the Health Sciences, Befhesda, MD.
Busvine, J. R. 1964. The significance of DDT-irritability
tests on mosquitoes. Bull. WHO 31:645-656.
Chareonviriyaphap, T, D. R. Roberts, R. G. Andre, H. J.
Harlan, S. Manguin and M. J. Bangs. 1997. Pesticide
avoidance behavior in Anopheles albimanus, a malaria
vector of Central and South America. J. Am. Mosq.
Control Assoc. 13 (in press).
Charlwood, J. D. and N. D. Paraluppi. 1978. The use of
excito-repellency boxes with Anopheles darlingi Root,
A. nuneztovari Gabaldon and Culex pipiens quinque-
fasciatus Say, obtained from areas near Manaus, Ama-
zonas. Acta Amazonica 8:605-611.
Collett, D. 1994. Modelling survival data in medical re-
search. Chapman & Hall, New York.
Coluzzi, M. 1963. Studies on irritability of DDT to
anopheline mosquitoes. WHO/VBC 33:1-22.
Elliott, R. 1972. The influence of vector behavior on ma-
laria transmission. Am. J. Trap. Med. Hyg. 21:755-763.
Evans, R. G. 1993. Laboratory evaluation of the irritancy
of bendiocarb, lambda-cyhalothrin and DDT to Anoph-
eles gambiae. J. Am. Mosq. Control Assoc. 9:285-293.
Lee, E. T. 1992. Statistical methods for survival data
analysis. John Wiley & Sons, Inc., New York.
Mantel, N. and W. Haenzel. 1959. Statistical aspects of
the analysis of data from retrospective studies of dis-
eases. J. Natl. Cancer Inst. 22:719-748.
Muirhead-Thomson, R. C. 1960. The significance of ir-
ritability, behavioristic avoidance and allied phenomena
in malaria eradication. Bull. WHO 22:721-734.
Rachou, R. G., L. A. Schinazi and M. Moura Lima. 1973.
An intensive study of the causes for the failure of re-
sidual DDT spraying to interrupt the transmission of
malaria in Atalaya and Falla, two villages on the coastal
plain of El Salvador, Central America. Rev. Bras. Ma-
lariol. Doencas Trap. 25:5-293.
Roberts, D. R. and W. D. Alecrim. 1991. Behavioral re-
sponse of Anopheles darlingi to DDT-sprayed house
walls in Amazonia. Bull. PAHO 25:210-217.
Roberts, D. R. and R. G. Andre. 1994. Insecticide resis-
tance issues in vector-borne disease control. Am. J.
Trop. Med. Hyg. 50(6)(Suppl.):21-34.
Roberts, D. R., W. D. Alecrim, A. M. Tavares and K. M.
McNeill. 1984. Influence of physiological condition on
the behavioral response of Anopheles darlingi to DDT.
Mosq. News 44:357-362.
Rozendaal, J. A., J. P. M. Van Hoof, J. Voorham and B.
F. J. Oostburg. 1989. Behavioral studies of Anopheles
darlingi in Suriname to DDT residues on house walls.
J. Am. Mosq. Control Assoc. 5:339-350.
Shalaby, A. M. 1966. Observations on some responses
of Anopheles culicifacies to DDT in experimental huts
in Gujarat State, India. Ann. Entomol. Soc. Am. 59:
Smith, A. and D. J. Webley. 1968. A verandah-type hut
for studying the house-frequenting habits of mosquitoes
and for assessing insecticides. III. The effect of DDT
on behaviour and mortality. Bull. Entomol. Res. 59:33-
World Health Organization. 1975. Manual on practical
entomology in malaria. Part II, pp. 157-163. World
Health Organization, Geneva, Switzerland.
J.S. Cohen (MD)
T.J. Gallagher (VA)
J.L. Heusel (GA)
R.G. Neumann (DE)
MID-ATLANTIC MOSQUITO CONTROL ASSOCIATION
-RICHLAND COUNTY, 2000 HAMPTON STREET, COLUMBIA, SC 29204
President: L.J. Bohn (Chesapeake, VA)
Vice-President: R.J. Wolfe (Milford, DE)
Secretary/Treasurer: SC. Ferguson (Columbia, SC)
Northeast Vector Management
Quality Unlimited Products
Running on DC
Clarke Mosquito Control Products
Coastal Vector Control
22nd Annual Meeting will be held March 12-14, 1997 in Dover, Delaware
DM. Rosenberg (WV)
J.A. Strickhouser (NC)
R.M. Turner (SC)
J.R. O'Neill (Industry)
Wm. F. Strickhouser Company